Optimization of a Working State of a System

Abstract

A system for optimizing a dynamic system including controllable working units. The system includes a processor of a controller node connected to a plurality of working units; a memory on which are stored machine-readable instructions that when executed by the processor, cause the processor to: acquire real-time response coefficients data from the plurality of the working units associated with corresponding working points, construct a response matrix based on response coefficients data, generate a tangent plane at each of the working points based on the associated working unit response coefficient, and determine an optimal working point in an area adjacent to the working point within the tangent plane for each of the plurality of the working units.

What is claimed is: 
     
         1 . A system, comprising:
 a processor of a controller node connected to a plurality of working units;   a memory on which are stored machine-readable instructions that when executed by the processor, cause the processor to:   (i) acquire real-time response coefficients data from the plurality of the working units associated with corresponding working points;   (ii) construct a response matrix based on response coefficients data;   (iii) generate a tangent plane at each of the working points based on the associated working unit response coefficient; and   (iv) determine an optimal working point in an area adjacent to the working point within the tangent plane for each of the plurality of the working units.   
     
     
         2 . The system of  claim 1 , wherein the instructions further cause the processor to perform iterations of steps (ii)-(iv) for reaching an optimum working point. 
     
     
         3 . The system of  claim 1 , wherein the instructions further cause the processor to acquire the response coefficients by superimposing a periodic small amplitude excitation Am, at the corresponding working point. 
     
     
         4 . The system of  claim 2 , wherein the instructions further cause the processor to collect resource consumption e, data and an output a  1  and to calculate amplitudes Δe i  and Δa j  at a corresponding frequency by FFT spectrum analysis. 
     
     
         5 . The system of  claim 4 , wherein the instructions further cause the processor to obtain resource consumption response coefficients and output response coefficients by dividing Δe i  and Δa j  by Δw i  respectively, wherein resource consumption response coefficient ρ i =Δe i /Δw i  and the output response coefficient c ij =Δa j /Δw i . 
     
     
         6 . The system of  claim 1 , wherein the response coefficient is mathematically equivalent to a partial derivative of an output function at the working point. 
     
     
         7 . The system of  claim 3 , wherein the amplitude excitation Δw i  is realized by periodically adding or subtracting a small amount Δw i  on w i , or by periodically changing its duty cycle by a PWM method. 
     
     
         8 . The system of  claim 1 , wherein the response matrix is composed of response coefficients is configured to express sensitivity of an output at the working point to a change of a working state. 
     
     
         9 . The system of  claim 1 , wherein the tangent plane is a tangent plane at the working point of a working surface in a multi-dimensional space, wherein an optimization process at the working point is transformed into a linear programming process in the tangent plane. 
     
     
         10 . The system of  claim 1 , wherein the adjacent area is an adjacent area of the working point w (k)  in the tangent plane, wherein a range of the adjacent area Δw (k+1)  ensures that any value in the adjacent area satisfies a condition w k+1) =w (k) +Δw k+1)  in a region w. 
     
     
         11 . The system of  claim 2 , wherein if during the iteration resource consumption changes in reverse, a range of a neighboring area is reduced to approach an optimal working point in smaller steps, wherein the iteration comprises going back two steps to start from a previous working point w (k+1)  for subsequent iterations. 
     
     
         12 . The system of  claim 2 , wherein a termination condition for the iteration is any of:
 a difference between resource consumption of two consecutive working points w (k)  and w (k+1)  is less than a set value; and   a step size between two consecutive operating points w (k)  and w (k+1)  is less than a set value.   
     
     
         13 . A method, comprising:
 acquiring, by a controller node, real-time response coefficients data from a plurality of the working units associated with corresponding working points;   (ii) constructing, by the controller node, a response matrix based on response coefficients data;   (iii) generating, by the controller node, a tangent plane at each of the working points based on the associated working unit response coefficient; and   (iv) determining an optimal working point in an area adjacent to the working point within the tangent plane for each of the plurality of the working units.   
     
     
         14 . The method of  claim 13 , further comprising performing iterations of steps (ii)-(iv) for reaching an optimum working point. 
     
     
         15 . The method of  claim 13 , further comprising acquiring the response coefficients by superimposing a periodic small amplitude excitation Δw i  at the corresponding working point. 
     
     
         16 . The method of  claim 15 , further comprising collecting resource consumption e, data and an output a j  data and calculating amplitudes Δe i  and Δa j  at a corresponding frequency by FFT spectrum analysis. 
     
     
         17 . The method of  claim 16 , further comprising obtaining resource consumption response coefficients and output response coefficients by dividing Δe i , and Δa j  by Δw i  respectively, wherein resource consumption response coefficient ρ i =Δe i /Δw i  and the output response coefficient c i,j =Δa j /Δw i . 
     
     
         18 . A non-transitory computer readable medium comprising instructions, that when read by a processor, cause the processor to perform:
 acquiring real-time response coefficients data from a plurality of the working units associated with corresponding working points;   constructing a response matrix based on response coefficients data;   generating a tangent plane at each of the working points based on the associated working unit response coefficient; and   determining an optimal working point in an area adjacent to the working point within the tangent plane for each of the plurality of the working units.   
     
     
         19 . The non-transitory computer readable medium of  claim 18 , further comprising instructions, that when read by the processor, cause the processor to collect resource consumption e, data and an output a j  data and calculating amplitudes Δe i , and Δa j  at a corresponding frequency by FFT spectrum analysis. 
     
     
         20 . The non-transitory computer readable medium of  claim 19 , further comprising instructions, that when read by the processor, cause the processor to obtain resource consumption response coefficients and output response coefficients by dividing Δe i  and Δa j  by Δw i  respectively, wherein resource consumption response coefficient ρ i =Δe i /Δw i  and the output response coefficient c i,j =Δa j /Δw i .